KR100453790B1 - Techniques for multitrack positioning and controlling error growth in self-servowriting systems - Google Patents

Abstract

Techniques are provided that include multitrack positioning during self-servowriting on a storage medium and methods and systems for controlling error augmentation as this servo recording progresses step by step on the storage medium. The readback amplitudes of multiple bursts from previously recorded tracks are combined using parabolic interpolation relationships for positioning in burst recording on subsequent tracks. This technique is particularly useful for systems in which the servo recording deviates from the recording element in a stepwise direction on the medium. Also disclosed is a control technique for error augmentation that induces and stores a reference waveform for use in recording subsequent tracks. Individual reference adjustments from the multitrack positioning signal are combined into a weighted sum that controls the error amplification when the weight is properly calculated, as described herein.

Description

TECHNIQUES FOR MULTITRACK POSITIONING AND CONTROLLING ERROR GROWTH IN SELF-SERVOWRITING SYSTEMS}

TECHNICAL FIELD The present invention relates to a data storage medium, and more particularly, to self-servowriting a servo pattern on a storage medium.

Self-servo recording has become an attractive technique for generating servo patterns on disk files because it eliminates the need for expensive external positioning systems and can be executed outside the indoor atmosphere. In general, this technique involves using a read and write element that is installed on an actuator in situ of a disk drive to initially write a servo pattern and then use it to accurately position the actuator during drive operation by the user. Include.

Self-propagation techniques of both radial positioning servo patterns and circumferential timing patterns have recently been developed. For example, in commonly assigned US Pat. No. 5,659,436 (referenced herein) entitled "Forming a Radial Magnetic Propagation Pattern for Disc File Servo Recording," it is used to control the head position while recording the next servo track. The servo position signal being derived is derived from the readback amplitude of a single track recorded one step before. However, in the current disk file, the read element can be offset by several tracks from the write element on the actuator. As the offset of the recording element increases in this reading, the commonly assigned US Patent No. 5,757,574 (referenced here) entitled "Self-Servo Recording Method and System Including Maintaining Reference Levels within Available Dynamic Range" As described, it is desirable to use a combination of readback amplitudes from several already recorded tracks to provide a position signal for the next track. In this case, the track immediately before the track being recorded is inaccessible because of the offset between the read and write elements. The result of this process is that whenever several weighting coefficients are applied to their readback amplitude in a weighted summation, any track contributes to servo positioning on several subsequent tracks.

The main point in radial magnetic propagation is to control the increase in track shape error as the actuator progresses stepwise on the disk surface to record sequential servo pattern tracks. However, while the above described technique compensates for the read-write element offset, no method of controlling the track shape error increase has been proposed.

According to the present invention, a technique for controlling error augmentation as the servo recording proceeds step by step along the storage medium is disclosed with a multitrack positioning technique.

In this regard, the present invention is a method of servo recording on a data storage medium in which the center of the recording element is deviated from the center of the reading element along the direction in which the servo recording proceeds step by step. One or more bursts may be recorded on one track of the storage medium while servoning using a position signal derived from each readback amplitude of another burst already recorded on the plurality of tracks. The reference waveform is derived as a function of the position error waveform. The position error waveform corresponds to one or more position errors of the reading element in relation to other bursts. The reference waveform is used to record subsequent tracks on the storage medium when the readout element overlaps one track.

The reference waveform, in one embodiment, calculates at least one composite coefficient of the Discrete Fourier Transform of the position waveform, multiplies the composite coefficient by the composite filter coefficient f, thereby forming at least one filtered coefficient, and at least one filter By calculating the inverse discrete Fourier transform of the calculated coefficients and adding the inverse discrete Fourier transform to the nominal mean reference level to form a reference waveform. The filter coefficient f can be calculated from a predetermined function C of the closed loop response of the servo loop used for servo recording. Multiple reference waveforms from each track are combined into a weighted sum, and this weight is calculated according to the relative sensitivity of the position signal to shift the position of each track.

In another aspect of the present invention, which may be used separately or in connection with the first aspect described above, a data storage medium in which the center of the recording element is at least partially away from the center of the reading element along the direction in which the servo recording proceeds step by step. A method for servo recording is provided. In this multitrack positioning embodiment, one or more bursts are recorded on one track of the storage medium while servoning using a position signal derived from each readback amplitude of another burst already recorded on the plurality of tracks. In this embodiment, the position signal is derived using parabolic interpolation of the readback amplitudes of different bursts already recorded on the plurality of tracks.

In one embodiment, three tracks are used for parabolic interpolation, the center of which has the highest readback amplitude, the previous track of the center track with the lower amplitude and the subsequent track of the center track with the other lower amplitude. Certain forms of parabolic interpolation functions are also described herein.

1 shows a data storage medium having a storage medium and associated servo electronics for use in self-servo recording of the invention.

FIG. 2 is a partial view of the storage medium of FIG. 1 showing an example of a track described herein and a self-servo recording burst. FIG.

3 is a schematic diagram of a parabolic interpolation function superimposed with the readback amplitude of a burst from five tracks already recorded according to the present invention.

4 is a schematic diagram of potential nonlinearity of actual versus interpolated head position.

5 is a flow diagram of a technique for controlling an increase in error of the present invention.

<Brief description of the main parts of the drawing>

10: data storage system

12: disk file

14: electronic device

16: Voice Coil Motor ("VCM")

18: head

20: medium

22: processor

24: pattern generator

26: amplitude signal

32: VCM driver circuit

The subject matter of the present invention has been specifically pointed out and specifically claimed at the end of the specification. However, the present invention, together with other objects and advantages thereof, will be better understood with reference to the following detailed description of the preferred embodiment and the accompanying drawings.

1 is an element example of the data storage system 10, used for radial magnetic propagation and servo pattern recording. The disk file 12 is connected to the electronic device 14 to read and write patterns on the medium 20 and to move the actuator at the end of the head 18 almost radially across the medium 20. Activate the voice coil motor ("VCM") 16. The processor 22 controls the pattern generator 24 to record the pattern of magnetic transformation on the selected media area. The RF readback signal from the readout element is demodulated to produce an amplitude signal 26 that reflects the superposition of the already recorded magnetic shift pattern and the readout element. The amplitude signal is an analog-to-digital converter 28 (" A / D "). Are digitized and analyzed by the processor 22 to obtain a position signal. The processor 22 calculates a numerical control signal which is converted into analog form by a digital-to-analog converter ("DAC") and processed by the control current by the VCM driver circuit 32 to drive the VCM 16 and the head ( 18) is appropriately positioned.

FIG. 2 not only divides each track into multiple sectors, but also multiple propagation tracks 100 when the first sector 116 normally comes immediately after the disk rotation index, as determined by the index pulse from the disk spindle motor driver. , 101, 102, etc.) is a part of the recording medium for explaining the division into ( Each sector is reserved for use in the region 117 containing amplitude bursts for propagation and for precision timing propagation systems and for recording actual product servo patterns including sector ID fields and amplitude bursts or phase encoded patterns. It is further divided into regions 118. In one embodiment of this system, the propagation burst area 117 is over-recorded with user data during user operation, following self-servo recording. All regions 118 except for the portion containing the product servo pattern may be overwritten as user data.

Each propagation burst area is further divided into a plurality of burst slots in which an amplitude burst pattern for propagation is recorded. In this example, eight slots numbered 0-7 are shown. Also shown at the exemplary location on the medium is a read element 200 and a write element 202. The recording element is positioned relative to the recording track 105, and because the offset is large, the reading element spans several tracks already recorded. At the servo track interval of half the data cylinder interval, the readout element can overlap three tracks at any time, as shown in FIG.

In this figure, the diagonaled burst represents the burst on the already recorded track. (If the read / write offset is large, some tracks must be prepared before the servo based on multiple tracks can be used. A method of preparing this initial set of tracks is "compliant to control head movement. There is a variety of transfers, including those described in the US Patent Application, which is generally assigned to the art of "forming an initial set of tracks in a self-servo recording system using a compliant crashstop." It is assumed that a track set exists.)

Multitrack Position Signals :

In the multitrack servo mode of the present invention, all three readback amplitudes are used to calculate the servo position signal using parabolic interpolation. This is shown in FIG. 3 showing a demodulated readback signal with dots representing digitization values for three related time slots, and a parabola defined with three reads. The horizontal axis of this parabola is in the unit of the servo track and the slot number because the initial tracks are recorded at a desired interval.

The parabolic peak is located at position P and is given by

Where V _A , V _B , and V _C are the readback amplitudes of the three bursts. The center B burst is assumed to have the highest readback amplitude. A burst is recorded one step before B, and C burst is recorded one step after B. Equation 1 gives the position of the peak relative to the B track position, which is between -0.5 and +0.5. The servo position signal, PS, is equal to B track number plus P. In Fig. 3, for example, the B track number is 2, so the PS is about 2.3.

Because of the finite number of available time slots, the slot number returns to zero as the number of slots increases, so the relationship between tracks and slots is not always as simple as in this example. However, in N slots numbered from 0 to N-1, the slot number associated with any track is easily calculated since it is equal to N in the track number modulus.

Unlike the single burst servo mode, where the position signal represents the amplitude of the fraction (range = 0-1), the multiburst signal represents the number of interpolation tracks for the three active signal tracks that can be sensed by the reading element at any time. Thus, a change in PS 1.0 corresponds to one servo step. However, since the interpolated PS value is not a complete linear function of the head position, the differential sensitivity varies over the range of possible P values. Nonlinear portions are formed from the curvature of the leadback profile, while some are formed from parabolic approximation. 4 shows a plot of PS versus head position for a typical head.

The shape of this curve depends on the read width, the write width and the servo track spacing. In writing a new servo track, the head is always stepped in the forward direction in units of one step equal to the periodicity of the nonlinear curve. Thus, nonlinearity does not directly affect track spacing.

The next track to be recorded is located before the current one and may be track number 105, for example for the case shown in FIG. This corresponds to having an offset of recording in the same read as 2.7 because the recording element is located at track number 5 and the reading element is located at 2.3. In this case, the absolute interval between tracks is equal to the offset interval of recording in the read divided by 2.7.

Stepping in the forward direction is accomplished by changing the reference signal input to the servo. The position error signal, or PES, is equal to the reference signal minus PS, and the controller acts to change the VCM current to reduce this error to zero. Adding 1.0 to the reference signal causes the servo to reposition the head, thus increasing the PS by the same amount. After setting to the new position, the next track is recorded.

In some cases, especially in the case of rotary actuators, the offset of the recording in the reading will be changed as the actuator moves in an arc shape across the disc. In order to prevent a change in the absolute interval of the tracks, the point P at which recording should occur must be adjusted. This gradual change is to stop every 40 tracks in a manner similar to that described in US Pat. No. 5,659,436, or to specify the cylinder-to-cylinder spacing of the last few tracks and to make minor adjustments to the servo reference signal to maintain the desired spacing. Can be processed. If this interval is too large, the criterion will be slightly reduced. This corresponds to moving the readout element backward in relation to the propagation echo, so that the next recorded track can be close to the present one. The reference increment of the subsequent step remains exactly 1.0, but the track spacing is reduced.

PS nonlinearity has a direct impact on the open loop gain of the servo. It is desirable to compensate for maintaining servo stability if the offset changes during propagation. Also, as desired, appropriate control of error augmentation involves computations that depend on the closed loop response of the servo, so it is desirable to keep it nearly constant. This can be accomplished by adjusting the servo gain to a predetermined coefficient based on the measurement of the position nonlinearity curve using an external positioning device for the representative disk file.

Alternatively, the measurement of the closed loop response can be performed by applying a sine wave modulated signal to the servo reference signal and measuring the amplitude and phase of the position error signal or the final modulation of the PES. The closed loop response is equal to one minus the ratio of PES to applied reference modulation. Each time a fractional servo point P is changed to maintain a constant track interval, it adjusts the closed loop response and adjusts the servo gain until a sufficiently close match is obtained. Alternatively, the servo gain can be determined at the start of propagation at several representative fractional servo points and interpolated to obtain new gains over the propagation itself.

In practice, the transfer function only needs to be measured at a single frequency. A good choice is when the magnitude of the closed loop response is approximately 0.5 (typically 10-15 times the rotational frequency). This frequency is large enough to have very little impact, such as actuator pivot characteristics, and low enough that the servo response can be measured quickly and accurately. It is also desirable to prevent large resonances, such as the butterfly mode of the actuator. With the choice of a good frequency, the magnitude of the open or closed loop response changes almost directly in proportion to the overall gain factor, making the iterative adjustment of the gain a quick and simple process.

Control of track shape error increase :

In self-servo recording, the track shape error is moved forward from step to step because the servo tracks the error for the current track when the servo writes a new one. The track shape error acts like an additional unintended reference signal input to the servo such that the response is transformed through the closed loop response of the system. Since the errors recorded therein can be repeated with disk rotation, they can be represented using discrete Fourier transforms with coefficients from integer multiples of the rotation frequency to a maximum frequency multiple equal to half the number of sectors.

Typically, a fairly strong control loop will be very close to 1 at low frequencies, exceeding 1 at intermediate frequencies, and then have a closed loop response that drops to zero at high frequencies. If some type of error correction is not applied, frequency components whose closed loop response exceeds 1 will rise exponentially with the number of steps.

In US Pat. No. 5,659,436, the control of track shape error enhancement calculates the Discrete Fourier Transform of the PES while recording the track, multiplies this coefficient by a vector of complex filter coefficients, and obtains a time domain waveform of an AC reference correction value. It includes modifying. The AC reference correction value is added to the DC portion of the servo reference signal (also referred to as the nominal average reference level) and used after stepping to the track just recorded. The filter coefficients are computed using the formula f = (SC) / (1-C), where C is a vector of complex values equal to the closed loop response of the servo at integer multiples of the disk rotation frequency and S is the error of the step in the step. Magnification factor. If S has a size smaller than one unit, the error is reduced and the propagation process becomes stable.

In the present invention, this technique is extended to cover multitrack servo processing as described above. This reference adjustment is calculated as before for each recording track and stored for use when the reading element which actually delays recording by several steps actually reaches the track. Since the multitrack servo process involves three recording tracks at a time in this example embodiment, the individual reference adjustments can be combined into a weighted sum.

More specifically, the method of the present invention includes the following steps with reference to FIG.

Servo control is performed at position X defined by the readback amplitude from the three tracks already recorded (step 510).

A new track number W is recorded by enabling the recording of the radial burst pattern for the next available time slot (the number of W modulo slots). During writing, the servo PES is written for each sector, resulting in a discrete time domain waveform stored in memory (step 520).

The waveform of the PES value is transformed using a Discrete Fourier Transform or a DFT to obtain a set of complex frequency domain coefficients (step 530). They are multiplied by a vector of complex filter coefficients f that have already been computed according to equation f = (S-C) / (1-C), where C is the closed loop response of the servo and S is a number less than one (step 540). The scaled coefficient is used to implement an inverse DFT which results in a discrete time domain waveform of the reference correction value R. They are indexed with the recorded track number W and sector s, ie R (W, s), and stored in memory for later use.

The head is advanced in steps forward by one track by changing the servo reference. The servo reference signal for each sector is set equal to the DC reference signal plus the AC reference signal. The DC criterion is X + 1.0 and is the same for every sector. The AC reference signal is the sum of three items, w _A R (t _A , s) + w _B R (t _B , s) + w _C R (t _C , s). Where A, B and C represent the roles played by each track in parabolic approximation, and w _A , w _B and w _C are the weighting coefficients associated with each role. R are reference correction values already stored in the corresponding track, t _A , t _B , t _C and sector s (step 560).

After the head is normally set at a new track position, which is one rotation of the disc later, a new track W + 1 is recorded and the process is repeated.

Proper selection of the reference correction weighting coefficient is very important for controlling the error augmentation. The present invention provides a method for ensuring that errors are eliminated. The concept below is that the weight reflects the relative contribution of the track position error to the PES. Assuming the deviation is small, the position signal difference is a derivative, δp = Is obtained by applying a chain rule to. Here, δP refers to a position signal change resulting from the change of the track position δX. Since the derivative depends on the burst (A, B, C) being considered, this needs to be analyzed for all three. For parabolic interpolation, Equation 1 is for each Can be differentiated to give.

The second link in the derivative chain, Is the derivative of the readback profile. Since this can vary from head to head, it is best to actually measure during or at the start of the servolite process. If the offset of the read in the write is changing, this can be done during the recalibration immediately following the determination of the new DC servo reference value. The measurement can be performed as follows. During servo to position P + ΔP, three readback amplitudes are recorded. The amplitude is again measured at position P-ΔP and subtracted from the first reading. ΔP is a position change as small as 0.05, for example. This provides a voltage change of PS change that is 2ΔP. The voltage derivative is equal to BB, where ΔV represents the difference in readback amplitude. Three weighting factors are given by:

The parabola is one of many possible ways of calculating the interpolated position signal from the readback amplitude on multiple tracks. The present invention can be applied to any technique in which the position signal depends on a plurality of readback amplitudes. The waveform of the reference correction value is calculated as described above and stored for each recording track. These are combined using a weighted sum to obtain an AC reference correction applied to the servo, shifting the position of each related track respectively by making this weight equal to the relative sensitivity of the position signal. These are the same as the partial derivative of the position signal in relation to the readback amplitude (which can be derived from a specific parabola) multiplied by the derivative of the readback amplitude in relation to the head position (which can be measured as described above).

The filter coefficient used to compute the stored reference waveform for each recording track depends on the closed loop response of the servo. Typically, this remains almost constant throughout the propagation, so that only what is determined at the start of the propagation is necessary, or may be predetermined based on measurements on a representative disk file. Large changes in actuator operation, such as when the arm encounters obstacles such as load / unload ramps, can cause a significant change in the closed loop response. Normally this causes the track shape error to increase quickly, which provides a very sensitive detection mechanism.

While the invention has been particularly shown and described in connection with the preferred embodiments, it will be understood by those skilled in the art that various modifications and details can be made without departing from the spirit and scope of the invention.

The present invention provides a method and system for controlling multitrack positioning during self-servo recording on a storage medium and error augmentation as the servo recording progresses step by step on the storage medium.

Claims (16)

1. A method for servo recording on a data storage medium of a data storage device which is generally out of the center of the reading element along the direction in which servo writing proceeds step by step, Servo recording one or more bursts on one track of the storage medium using position signals derived from each readback amplitude of another burst already recorded on the plurality of tracks; Deriving a reference waveform as a function of a position error waveform, the position error waveform corresponding to one or more position errors of the readout element related to the other burst; Combining a plurality of reference waveforms corresponding to the tracks used to derive the position signal to provide a servo reference waveform, and when the readout element overlaps the one track, the storage medium is utilized using the servo reference waveform. Recording subsequent tracks on the screen Servo recording method comprising a. The method of claim 1, The derivation step, Calculating one or more complex coefficients of a discrete Fourier transform of the position error waveform; Multiplying the one or more composite coefficients by one or more composite filter coefficients f to produce one or more filtered coefficients; Calculating an inverse discrete Fourier transform from the one or more filtered coefficients, Adding the inverse discrete Fourier transform to a nominal average reference level to form the reference waveform Servo recording method comprising a. 1. A method for servo recording on a data storage medium of a data storage device, generally out of the center of the reading element, in the direction in which the servo writing proceeds step by step, Recording one or more bursts on one track of the storage medium while servoning using a position signal derived from each readback amplitude of another burst already recorded on a plurality of tracks. Including, And the position signal is derived using parabolic interpolation of the readback amplitude of the other burst already recorded on the plurality of tracks. A system for servo recording on a data storage medium of a data storage device which is generally out of the center of the reading element along the direction in which the servo writing proceeds step by step, Means for servo-recording one or more bursts on one track of the storage medium using position signals derived from each readback amplitude of another burst already recorded on a plurality of tracks; Means for deriving a reference waveform as a function of a position error waveform, the position error waveform corresponding to one or more position errors of the reading element relative to the other burst; Means for providing a servo reference waveform by combining a plurality of reference waveforms corresponding to the tracks used to derive the position signal, and the storage using the reference waveform when the readout element overlaps the one track. Means for recording subsequent tracks on the medium Servo recording system comprising a. The method of claim 4, wherein The induction means, Means for calculating one or more complex coefficients of a discrete Fourier transform of the position error waveform; Means for multiplying the one or more composite coefficients with one or more composite filter coefficients f to produce one or more filtered coefficients; Means for calculating an inverse discrete Fourier transform from the one or more filtered coefficients, Means for adding the inverse discrete Fourier transform to a nominal average reference level to form the reference waveform Servo recording system comprising a. The method of claim 5, And means for calculating f from a predetermined function C of the closed loop response of the servo loop used for the servo recording. The method of claim 6, Means for calculating f comprises means for using the relationship f = (S-C) / (1-C), where S is a step coefficient. The method of claim 4, wherein And the position signal is derived using parabolic interpolation of the readback amplitude of the other burst already recorded on the plurality of tracks. The method of claim 8, And the plurality of tracks comprises a center track where the read elements intersect and has the highest readback amplitude, a previous track of the center track, and a subsequent track of the center track. The method of claim 4, wherein And the stored reference waveforms corresponding to the tracks used in deriving the position signal are combined in a weighted sum to provide the reference waveforms used in recording the subsequent tracks. The method of claim 10, And shift each position of each track by equalizing each weight of the weighted sum equal to the relative sensitivity of the position signal. The method of claim 11, And the relative sensitivity is obtained by multiplying the partial derivative of the position signal according to the readback amplitude by the derivative of the readback amplitude according to the position of each track. The method of claim 4, wherein And said center of said recording element is deviated by an amount greater than that for the recorded track width from said center of said reading element. In general, a system for servo recording on a data storage medium of a data storage device which is off the center of the recording element at the center of the reading element along the direction in which the servo recording proceeds step by step, Means for recording, while servo-serving, one or more bursts on one track of the storage medium using a position signal derived from each readback amplitude of another burst already recorded on a plurality of tracks; Means for deriving the position signal using parabolic interpolation of the readback amplitude of the other burst already recorded on the plurality of tracks Servo recording system comprising a. The method of claim 14, The plurality of tracks include a center track intersecting the reading element and having the highest readback amplitude ("V _B "), and a track before the center track having each readback amplitude ("V _A "). And a track subsequent to said center track with each readback amplitude ("V _C "). The method of claim 15, The parabolic interpolation is substantially Servo recording system characterized in that the form of.